Over the weekend, my LinkedIn feed was flooded with photos of individuals taking selfies with everybody from Bollywood actor Shah Rukh Khan to Tesla boss Elon Musk. In a single picture, an govt gave the impression to be taking a selfie with Shah Rukh Khan at Harvard Enterprise Faculty, and the picture seemed utterly plausible.
Nevertheless, once I learn the publish, I found that the image had been created utilizing Nano Banana, a brand new picture enhancing mannequin from Google’s DeepMind unit. The mannequin was first launched final week, and later built-in into Google Gemini.
I attempted it, and right here’s why Google’s new picture enhancing mannequin goes viral on social media (for all the precise causes).
Find out how to use Nano Banana
Google’s Gemini can already create AI-generated photographs, however the outcomes haven’t at all times been probably the most spectacular. Google DeepMind unit’s new AI mannequin, nicknamed “Nano Banana” and formally referred to as Gemini 2.5 Flash Picture, works equally to most AI picture enhancing instruments: you add a picture and describe the adjustments you need.
What units this mannequin aside, nonetheless, is its potential to deal with extremely particular edits, from altering a topic’s garments or location to including one other individual (like a celeb) right into a selfie. It’s enjoyable to make use of Google’s Nano Banana mannequin. To assist forestall confusion over whether or not photographs are actual or AI-generated, all outputs embrace each a visual watermark and an invisible one, referred to as SynthID.
All photographs have a visual watermark and an invisible watermark referred to as SynthID to clear any confusion over whether or not they’re actual or AI-generated.
Enhancing with Nano Banana
To edit a picture utilizing Nano Banana, open Google AI Studio and register along with your Gmail account. Alternatively, you should use Google Gemini. When you open it, you will notice a banner on the prime that claims: NEW! Attempt picture enhancing with our greatest picture mannequin, Nano Banana. From there, merely add a picture, write a immediate, and your edited picture will likely be prepared in just a few seconds.
Selfie with Shah Rukh Khan
I attempted enhancing a picture utilizing a number of completely different situations. The very first thing I imagined was taking a selfie with Shah Rukh Khan. For instance, I used a current selfie I took in China and wrote the immediate: “A selfie with Shah Rukh Khan.” In only a few seconds, I acquired the end result, and I used to be shocked at how practical it seemed. It confirmed me standing subsequent to Shah Rukh Khan, and the picture was convincing.
Story continues under this advert
Once I shared the AI-generated image in a WhatsApp group, one in every of my colleagues even gave it a thumbs-up, considering I had truly taken a selfie with him.
Right here’s the unique picture (left), and right here’s the Gemini-generated picture, a selfie with Shah Rukh Khan (proper).
Should you look intently, you’ll discover that the AI didn’t de-age Khan an excessive amount of, in contrast to in a few of his current motion pictures, reminiscent of Pathaan and Jawan, the place he has been closely de-aged utilizing CGI (as is widespread with many senior male actors today). He doesn’t appear to be he did in his late 20s, and that truly provides to the realism. The distinction in peak between us was additionally considered. Khan isn’t notably tall, and neither am I.
To check the mannequin’s accuracy, I attempted one other immediate: “A selfie with Brad Pitt.” When Gemini created that picture, the peak distinction between me and Pitt was clearly seen, additional proving how convincing the Nano Banana mannequin actually is.
One other selfie – this time with Brad Pitt. Are you able to belive it’s an AI generated picture?
Sporting a kimono on the streets of Tokyo
As quickly as I noticed the convincing selfie with Khan, my inventive juices began to movement. I then imagined myself in a kimono, an extended gown secured across the waist with an obi belt. It’s a conventional Japanese garment worn by women and men.
Story continues under this advert
Right here’s the immediate I used: “Make me put on a Kimono on the streets of Tokyo.”
The outcomes I acquired had been a blended bag. The primary image I uploaded was a partial selfie (not a full-body picture), and whereas Gemini did efficiently change the background ( imagining me sporting a kimono on the streets of Tokyo), the face within the picture wasn’t mine. It seemed AI-generated.
I’d say the individual within the Gemini-generated image resembled me considerably, but it surely wasn’t fairly proper. I imagine that to generate a extra correct picture, Gemini wants a full-length picture – ideally, a close-up image of mine. It stays a posh activity for AI to generate a sensible picture when requested to think about a complete situation from only a selfie, particularly when aiming for an correct illustration.
The Nano Banana AI mannequin does make errors, as you possibly can see within the picture on the precise. I don’t just like the individual wearing a kimono.
Digital try-on
One other use case I found for Gemini’s picture enhancing with the Nano Banana mannequin is attempting on digital garments. I uploaded a current image of myself, taken contained in the altering room of a Uniqlo retailer in Shenzhen, China, and wrote a immediate asking Gemini to decorate me in a go well with.
Story continues under this advert
Though I don’t normally put on fits. In truth, I really feel a bit nervous sporting them, however this enjoyable experiment helped me think about what I would appear to be if I ever wanted to put on one. Earlier than going to a retailer to strive on a ready-made go well with, I might merely add a photograph and ask Gemini to decorate me in a basic black go well with.
It’d sound a bit foolish, particularly since many elements contribute to creating a well-fitted go well with, however this type of digital try-on can enhance confidence, and enable you really feel higher about your self and your physique, simply as I did.
The outcomes had been spectacular. Gemini generated a picture that precisely represented my physique sort and peak. My face wasn’t pixelated, and the picture high quality wasn’t compromised in any manner. Nevertheless, I observed my watch was faraway from my wrist, so it does make errors. To make clear: I’m not as lean as I appeared within the Gemini-generated image, however that’s not the AI’s fault. The T-shirt I attempted on in Uniqlo’s trial room made me seem slimmer than I truly am.
Digital try-on is the spectacular use case of the newly launched Google AI mannequin.
Extra use circumstances to be explored
It’s nonetheless early days for Google’s Nano Banana mannequin, and I’m fairly certain that many extra use circumstances are but to be explored. Primarily based on my time testing the brand new AI mannequin, I can see the way it’s being skilled to make topics extra constant and correct, a step ahead for AI-generated photographs, which beforehand typically felt both unpredictable or removed from actuality – a minimum of the place people are concerned.